48Vdc to 220Vac 5KVA inverter - MOSFETs are buring

[masifshahzad]

Hi,
I have designed 5KVA inverter which is using 4 batteries in series giving 48V. I am using IRF640N mosfet which is 18A mosfet. My transformer is 48-0-48 so i have used 12 mosfets in parallel at one side so i am using 24 mosfets in total. But when i check my inverter on small transformer its working fine but i connect it with my 5KVA transformer it burn all the mosfets of one side (11 mosfets).i use 35A diodes at the input of 5KVA transformer. Could you please tell me why mosfets rae buring but only of oneside?

Added after 14 minutes:

i am generating 50Hz squarewave from CD4047 IC and after it i am using ICL7667 which is MOSFET driver.Could you please tell me whether the problem is with transformer or selection of MOSFETs. around more than 100amp or current will flow from one side of mosfets (12 mosfets) so each share less than 10amp.

Added after 6 minutes:

The circuit is working fine with small size transformer which is 12-0-12 to 220V of around 2 amp.
One important thing is that mosfets burn when i switched on the circuit and no load is connected at the output of 5KVA transformer.

 

[dick_freebird]

Are you using one driver per FET, or are you trying to run the
whole rack off one poor little driver?

Seems to me like most of the questions ought to be answered
with an oscilloscope rather than speculation (without even the
benefit of a schematic).

 

[masifshahzad]

yes i am using two ICs of ICL7667. each IC has two mosfets drivers. i have shorted the two drivers of each IC to drive 12 mosfets. so both the ICs with two drivers in parallel are driving 24 mosfets IRF640N.

Added after 4 minutes:

This happend to me twice that same side of MOSFETs have burned when i switch ON the circuit at NO LOAD.

 

[dick_freebird]

I think your driver is undersized. Also I see no mention of any
undervoltage lockout, so there seems to me a good chance of the
FET gates being driven to unknown voltages during power-up.

On the other hand if switching does not begin briskly, you could
have one rack of FETs sitting there conducting against nothing
but copper and battery. What do you know about how fast
the 4047 starts chopping?

Maybe you want to gate that output so that the FETs on both sides
are off initially, and only release them when you have a good
enough 12V rail, a running oscillator and so on.

 

[stanislas]

Use mosfet of higher amp and ur problem is solved

 

[mvs sarma]

insufficient Gate voltage also heats up the FTTs.
unwanted capacitance on the gate drives round off the pwm signals ,
perhaps a good MOSFET driver or check the present driver section
settles the issue.

One reason for one side MOSFETs burning can be interpreted to initial 12V present on the gates
for longer time before the oscillation and/or pwm can take over.
we may perhaps delay the Drain supply 48V, to the final stage by a second by that time the drive signal stabilizes. it might need a delay for the relay operation
at the same time , the supply to the earlier circuitry could be taken thro' an alternate switch or a switched transistor.

 

[masifshahzad]

Thank you so much for your replies.
OK let me redesign the mosfets driver circuit and check it again.May be the problem is with 4047 ic or the 7667 ic. I let u know by sending the new driver schematic.

One thing to ask is can i use mosfet as a pre-driver to drive these 2 sets of parallel mosfets. that pre-driver mosfet will be drived by the mosfet driver icl7667.????

Thanks.

 

[mvs sarma]

Thank you so much for your replies.
OK let me redesign the mosfets driver circuit and check it again.May be the problem is with 4047 ic or the 7667 ic. I let u know by sending the new driver schematic.

One thing to ask is can i use mosfet as a pre-driver to drive these 2 sets of parallel mosfets. that pre-driver mosfet will be drived by the mosfet driver icl7667.????

Thanks.

while that aspect has to be debated, i feel, it better to have fewer mosfets of higher capacity , rather than more mosfets as the gates would be common increasing the total capacitive load.

 

[FvM]

Weak gate drivers won't explain why the power stage is blown at no load. I fear, it may a problem of high inrush currents due to core saturation. Starting the inverter operation with a halfwave of 10 ms duration implies doubled flux compared to periodic operation. To avoid saturation, the first halfwave should be shortened to 5 ms.

Or use any other means to limit the inrush current, e.g. a current limiting resistor. Furthermore, it seems appropriate to have some kind of overcurrent shutdown for a high power inverter.

 

[kabiru]

Hi, I think your problem is from your oscilator cd4047 is not a PWM ic's I advice you use PWM ic's like SG3524 ,3525, TL494 or 7500-also your inverter can easily burn due to core saturation therefore you reduce the on time by reducing the duty cycle, if you do that your problem will be solved.

 

[masifshahzad]

Hi,

@ MVS SHARMA - You are right but problem is that high rating mosfets are usually not available in my area and if available then they are so much costly. IRF3205 is available, its Id is 110A but problem is with Vds which is just 55V which is quiet near to 48V. Can you suggest any high rating mosfet with high Id, Vds and low on resistance???

Added after 6 minutes:

Hi,

@ FVM - So many people suggest me the same solution of sudden inrush current. They said that mosfets are burning due to sudden inrush current. One suggests me to short the output or input winding of transformer after switching OFF the circuit to discharge the saturated core.

Current limiting resistor is a good solution but i would require resistor of high power rating.

I think this is also true that 10ms is huge time which saturates the core. This should be reduced.

Added after 1 minutes:

Hi,

@ KABIRU - Yes i m redesigning the PWM circuit using PIC microcontroller. This will solve the problem of huge ON time.

Thanks.

 

[FvM]

One suggests me to short the output or input winding of transformer after switching OFF the circuit to discharge the saturated core.

After power off, only a small remanent flux is kept in the transformer. It's polarity actually plays a certain role,
if the direction of the first halfwave is the same, saturation will happen more easily. But if you use a reasonable startup
sequence, either a PWM ramp or reduced duration of the first thalfwave, you don't have to worry about remanent
transformer magnetization.

Yes i m redesigning the PWM circuit using PIC microcontroller.

This way, you should be able to fix the problems. I suggest an overcurrent detection fed to the PIC as additional option.

 

[masifshahzad]

Yes u r right. Now i am designing a circuit which is totally microcontroller based. That controller will generate PWM, sense charging AC voltage, generated AC voltage, voltmeter for AC and DC and many more options. Soon i will discuss the schematic with you.

For overload protection, i think current transformer is best choice.

 

[mvs sarma]

Hi,

@ MVS SHARMA - You are right but problem is that high rating mosfets are usually not available in my area and if available then they are so much costly. IRF3205 is available, its Id is 110A but problem is with Vds which is just 55V which is quiet near to 48V. Can you suggest any high rating mosfet with high Id, Vds and low on resistance???

Added after 6 minutes:

Hi,

@ FVM - .......

Thanks.

Perhaps you need to have a minimum of double or little more Vds than the working voltage. The auxilary supply serving the electronics including gate drive should have adequate capacity to prevent under drive that could cause temperature raise.
thus I personally feel 150V, 40 to 50 amp devices would help you. You need only 4 on each side.

the FETs need to have gate termination instead of floating, unless the driver takes care of it. did i miss the schematic element of the inverter by any chance?
it really helps to comment more logically.

PS: Please see the application diagram Fig10B on the datasheet of ICL7667 where in the input pins are pulled up.(perhaps also due to open collector configuration of TL494.

in the test they clearly specify not to float the input pins.. perhaps a preferred practice could be use of High frequency isolation transformer so that only AC pulses can flow thro . this prevents dc voltage sittting on a fet set and possible killing them

 

[masifshahzad]

Yes sharma i havent posted the schematic yet. I am redesigning it, let me complete it. I will post it here soon for discussion.

Thanks

 

[massi]

Hi i'm massi i have similar problem with mosfet's,

Its like the ups works fine for some time may be after 2/3 months the mosfet bank
goes off... everytime in need to change this i tried different rating capacity but no use. please do help... i use Dsp 301series microcontroller to control the ups.
And i'm looking for some 3kva online ups its better if higher i run my comps 24/7

Thank u in advance
please do the need full:-(

 

[Orson Cart]

FVM is correct:

Weak gate drivers won't explain why the power stage is blown at no load. I fear, it may a problem of high inrush currents due to core saturation. Starting the inverter operation with a halfwave of 10 ms duration implies doubled flux compared to periodic operation. To avoid saturation, the first halfwave should be shortened to 5 ms.
Or use any other means to limit the inrush current, e.g. a current limiting resistor. Furthermore, it seems appropriate to have some kind of overcurrent shutdown for a high power inverter.

This is why the original circuit worked on a small transformer as the resistance of the windings was enough to limit overcurrents due to applying too many volt seconds to the transformer at startup, also if you are running a push pull on the transformer you need to allow a dead time for the flux to self reset in case the volt seconds you apply on each half cycle are not identical, simple push-pull transformer drive leads to flux "staircasing" and eventual overcurrent on one bank of fets due to transformer saturation, you will notice this more on a big transformer where the resistance is much lower, Regards, Orson Cart.

 

[smxx]

hi
when you use 5KVA transformer ,you must make dv/dt protecion for your switchs by a RC branch. you have large L di/dt and hi voltage shock.

---------- Post added at 09:29 ---------- Previous post was at 09:28 ----------

use SPWM technic for best result

 

[kabiru]

yes, spwm will be the best .